An image decoding method according to the present document comprises the steps of: deriving flag information indicating whether valid coefficients are present in a second area excluding a first area of the upper left corner of the current block; parsing an MTS index from a bitstream on the basis of the flag information indicating the absence of valid coefficients in the second area; and deriving residual samples for the current block by applying a conversion kernel derived on the basis of the MTS index to conversion coefficients of the first area, wherein the flag information may be derived by determining, for each scan sub-block in which the valid coefficients are scanned, whether the valid coefficients are present in the second area.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
2. The image decoding method of claim 1, wherein the MTS index is inferred to 0 when the MTS index is not parsed.
This invention relates to image decoding, specifically improving efficiency in handling motion texture synthesis (MTS) indices during the decoding process. The problem addressed is the computational overhead and complexity when decoding images that use MTS, particularly when the MTS index is not explicitly parsed from the encoded data. In such cases, the invention provides a default inference mechanism to set the MTS index to 0, ensuring smooth decoding without requiring additional parsing steps or error handling. The method involves decoding an image that includes motion texture synthesis data, where the MTS index is a parameter used to determine how texture synthesis is applied during motion compensation. If the encoded data does not include an explicit MTS index, the decoder automatically infers the index as 0, which simplifies the decoding process by eliminating the need to check for missing indices or handle undefined states. This approach reduces processing time and computational resources while maintaining decoding accuracy. The invention is particularly useful in video compression standards where MTS is employed to enhance motion compensation accuracy, such as in advanced video codecs. By standardizing the handling of missing MTS indices, the method ensures compatibility and efficiency across different decoding implementations. The solution is applicable to both hardware and software-based decoders, making it versatile for various applications in video streaming, broadcasting, and multimedia processing.
5. The image decoding method of claim 4, wherein the scan subblock is scanned according to the reverse diagonal scan order from a last scan subblock including a last significant coefficient in the current block.
This invention relates to image decoding, specifically improving the efficiency of scanning subblocks within a block of image data. The problem addressed is optimizing the scanning order of subblocks to reduce computational overhead and improve decoding performance, particularly in video compression standards like HEVC or AV1. The method involves scanning subblocks within a current block of image data in a reverse diagonal order. The scan starts from the last subblock that contains the last significant coefficient in the current block. This approach leverages the fact that significant coefficients (non-zero values) are often clustered in certain regions of the block, allowing the decoder to skip unnecessary subblocks more efficiently. By starting the scan from the last significant coefficient, the method minimizes the number of subblocks that need to be processed, reducing memory access and computational steps. The reverse diagonal scan order ensures that adjacent subblocks are scanned in a way that aligns with common coefficient distribution patterns, further optimizing the decoding process. This technique is particularly useful in high-efficiency video coding where subblock-based processing is common. The method improves decoding speed and reduces power consumption, making it suitable for real-time applications and resource-constrained devices.
7. The image encoding method of claim 6, wherein the MTS index is inferred to 0 when the MTS index is not encoded.
This invention relates to image encoding, specifically improving efficiency by selectively encoding a motion tree structure (MTS) index. The problem addressed is the computational and bandwidth overhead of encoding MTS indices in video compression, particularly when the index remains unchanged or follows predictable patterns. The solution involves inferring the MTS index as zero when it is not explicitly encoded, reducing redundancy without sacrificing accuracy. The method applies to video encoding systems where motion information is hierarchically organized, such as in tree-based motion partitioning schemes. By omitting the encoding of zero or default MTS indices, the technique minimizes bitrate while maintaining compatibility with existing decoding processes. The approach leverages context-based decisions to determine when index encoding is unnecessary, optimizing both encoding speed and compression efficiency. This is particularly useful in scenarios where motion consistency is high, such as in regions with uniform motion or static backgrounds. The invention builds on prior techniques for motion tree structure encoding but introduces a conditional encoding mechanism to skip redundant index transmissions. The method ensures backward compatibility by defining clear rules for index inference, allowing decoders to reconstruct the MTS index correctly even when it is not explicitly transmitted. This reduces the overall data payload while preserving the integrity of motion information in encoded video streams.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
November 9, 2020
April 16, 2024
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.